基于离散增量和协变判别函数识别蛋白质亚核定位

Identifing Subnuclear Localization by Using the Increment of Diversity Combined with the Covariant Discriminant Function

利用离散增量结合协变判别函数,选取氨基酸组份和N端氨基酸二肽组份为信息参数,对蛋白质亚核定位进行预测.在序列相似性小于等于25%时,406个单定位亚核蛋白Jackknife检验总预测成功率为75.9%,相关系数CC为0.644,把多定位亚核蛋白作为独立测试集,92个多定位亚核蛋白总预测成功率为78.3%.在序列相似性小于等于65%时,504个单定位亚核蛋白Jackknife检验总预测成功率为75.6%,相关系数CC为0.643,92个多定位亚核蛋白总预测成功率为80.4%.与Lei等人利用Lei-SVM方法对该数据库预测结果相比,单定位亚核蛋白总预测成功率比Lei等人高9.1%,CC值比Lei等人高0.124,多定位亚核蛋白总预测成功率比Lei等人高15.2%.

Based on the increment of diversity and covariant discriminant function,the protein subnuclear location was predicted by using of the amino acid compositions and N-terminal dipeptide compositions. For 406 single localization nuclear proteins with ≤25 % sequence identity, the results of jackknife test show that the overall accuracy and correlation coefficient are 75.9% and 0. 644, respectively. And the overall accuracy is 78. 3% for an independent set of 92 multi-localization proteins. For 504 single localization proteins in jackknife test the overall predictive success rate and correlation coefficient are 75.6% and 0. 643,respectively. And the overall correct predicted rate is 80.4% for the independent set of 92 multi-localization proteins. Our result is 9. 1% higher than the result of Lei＇s SVM for 504 single localization proteins,and 15.2% higher than Lei＇s SVM for an independent set of 92 multi-localization proteins.